Namespace IntentSharp.Providers
Classes
- AIAuthenticationException
HTTP 401 or 403 from the provider.
- AIGuardrailException
A limit (max model calls, max execution time) stopped the operation.
- AIMessage
A single chat message.
- AIProviderException
A provider call failed.
- AIProviderInternalException
HTTP 5xx from the provider.
- AIRateLimitException
HTTP 429 from the provider, or the local rate limiter refused the request.
- AIRequest
Provider-neutral request. The client builds one for each attempt; providers translate it to their wire format.
- AIResponse
Provider-neutral response.
- AITimeoutException
The request exceeded its timeout.
- AIValidationException
HTTP 400 or 422 from the provider: the request itself was rejected. HttpStatusCode is the actual status.
- AnthropicProvider
Anthropic Messages API provider. Supports JSON-schema structured output through
output_config.format. Anthropic does not offer an embeddings endpoint, so SupportsEmbeddings is false.
- HttpProviderBase
Shared plumbing for HTTP-based providers: request posting, status-code mapping, bounded retries with exponential backoff, and cancellation. Subclasses only translate to and from their wire format.
- HttpProviderSettings
Settings shared by the HTTP providers.
- IntentSharpException
Base exception for IntentSharp.
- OpenAIProvider
OpenAI Chat Completions provider. Supports JSON-schema structured output and text-embedding-3 embeddings.
- SemanticDatabaseException
The database runtime refused to execute a plan.
- TokenUsage
Token counts for one request.
Interfaces
- IAIProvider
A model provider. Implementations send one AIRequest and return one AIResponse. Providers are stateless beyond their HTTP client and are safe to share across threads.
Enums
- ResponseFormat
How the model must shape its reply.